We conducted staging observations on multiple clutches of purple sea urchin (Strongylocentrotus purpuratus) larvae from 2010–2014, fertilized and reared through settlement in the laboratory using our modifications of standard methods;  details below). In order to insure consistency of our staging scheme, we reared larvae in 5 different locations in North America (SEA-Seattle, WA; FHL-Friday Harbor Labs, WA; BML-Bodega Marine Laboratories, CA; HMS-Hopkins Marine Station, CA and UOG-Guelph, ON).
Urchins and spawning
Adult urchins derived from two distinct USA source populations: Slip Point (Clallam Bay, WA; SEA, FHL) and The Cultured Abalone Ltd (Goleta, CA; UOG, BML, HMS). At UOG, we maintained urchins in the Hagen Aqualab, University of Guelph, in artificial seawater (Instant Ocean™; Instant Ocean) at 12°C and 34 ppt salinity. We fed the urchins rehydrated kombu kelp (Laminaria sp.) ad libitum. For the SEA and FHL rearings, we used urchins maintained in subtidal cages suspended off the floating docks at FHL, fed throughout the year ad libitum with drift kelp (mainly blades of Nereocystis leutkeana).
For the HMS and BML rearings, we used urchins maintained in the dark in flow-through natural sea water tables at HMS, fed throughout the year ad libitum with giant kelp (Macrocystis pyrifera).
We spawned adult sea urchins by gentle shaking or intra-coelomic injection with 0.5 M KCl, and fertilized spawned eggs (>90% fertilization success) with diluted sperm using standard methods .
For all embryo and larval culturing in Guelph, we used 0.2 μm Millipore™-filtered artificial seawater (MFASW), sterilized with UV light. We cultured embryos at 14°C until hatching (24 hours) at which point we poured off swimming embryos and set up our main cultures at 1 larva/4 ml derived from one male and one female. We used a mechanical stirring system  to keep larvae in suspension, and fed them a mixture of Dunieliella tertiolecta (12 cells/μl) and Rhodomonas lens (6 cells/μl). We changed greater than 95% of the water every two days by reverse filtration, and fed the larvae as above.
For our embryo and larval rearings at FHL, SEA, HMS and BML, we derived our cultures from equal-part mixtures of single-parent fertilizations [either 3 males × 1 female (3M × 1F), 3F × 1M or 2F × 2M] at an initial density of 1 embryo/ml, and maintained either using a mechanical stirring system (FHL, HMS, BML; 3]) or a shaking water bath (SEA) to keep larvae in suspension. Starting on day 3, we fed larvae a mixture of Dunieliella teriolecta (3 cells/μl) and Rhodomonas spp. (2.5 cells/μl) every 2 days following water changes as described above. We reduced the larval density to 0.1-0.25 larvae/ml on about day 16 (6 arm plutei, before rudiment invagination). At FHL and BML, we cultured our larvae at sea table temperatures, which vary from an average of 10-14°C depending on the time of year. At HMS and SEA, we maintained our larvae at constant temperatures (14°C and 16°C, respectively).
To develop the staging scheme, we identified dozens of score-able soft tissue and skeletogenic juvenile characters (see also ), and characterized hundreds of total larva for the presence or absence of that character. Our goal was to identify characters whose relative timing of appearance would be consistent. As such, we ended up excluding several classes of juvenile characters that we identified as “heterochronic” in their appearance from larva-to-larva or batch-to-batch, notably, those outside of the rudiment proper (see Additional file 1: Figure S1). In the final staging scheme, we thus settled on characters that are defined by discrete morphological features (soft and hard structures) within the developing juvenile rudiment of S. purpuratus larvae, visible in live specimens either under differential interference contrast (DIC) or cross-polarized light. A detailed description of these stages is presented in the Results section and in Tables 1 and 2.
Our data on approximate numbers of days from fertilization to various stages (see third column in Tables 1 and 2) are compiled from our numerous fertilizations and rearings in different seasons and locations at 14ºC and under our various culturing conditions outlined above. We therefore believe that the results are robust with respect to population differences in different geographic locations, as well as water chemistry and other factors. We note that these timing data are truly approximations, as larvae in a given batch can vary substantially from these values, both in mean stage and in the variance among stages. For example, some batches of larvae, for reasons that we and others have not identified, undergo bouts of asexual larval budding [45, 91, 92], sometimes in a majority of the larvae within a culture vessel (data not shown). The resultant larval cultures are both delayed and more variable than larvae from a more typical rearing.
Stage length experimental design
To determine with more precision the lengths of individual skeletogenic stages, we conducted staging observations on individual larvae in two of our rearing locations. The majority of our staging data derive from larvae that we cultured at UOG in June 2011. We also report on our temperature comparison experiment conducted in SEA in September 2011. Throughout the Methods and Results sections, we refer to our ‘Guelph’ or ‘Seattle’ experiments, respectively.
For the Guelph staging studies, we selected 48 larvae on day 28 that had visible rudiments, and mounted them individually on a microscope slide with raised cover glass using modeling clay, in order to immobilize but not damage the larvae. We then used cross-polarized and DIC optics to stage each larva (0 hour time point) according to our staging scheme (see Results and Tables 1 & 2), photographed it (see below), and gently transferred it to an individual well in a 24 well plate (Costar 3524) with 1.5 ml of MFASW and algae at the same concentrations as in their rearing conditions (see above). We maintained the well plates for 24 hours at 14°C, at which point we again mounted and photographed each larva as above, and assigned it to a stage (24 hour time point). We continued to culture 24 of these larvae for an additional 24 hours in their same wells, and photographed and staged them one final time (48 hour time point).
The Seattle protocol was similar to that described above with the following differences: 1) the experiment began on day 21, so the experimental larvae only covered skeletogenic Stages 0–6; 2) we assigned 48 larvae at random to either a 12°C or 16°C treatment to examine temperature effects on staging progression; 3) we staged each larva at 0, 24, 48 and 96 hours of the experiment; and 4) we fed larvae for the duration of the experiment at the level at which they were fed throughout development (see above), with a full water change at 48 hours.
Stage length calculations
For both the Guelph and the Seattle datasets, we made the following three assumptions for the calculation of skeletogenic stage lengths (in hours) for each observed larva: 1) since larvae were observed at most once per day, larvae at each observation point were assumed to be at the mid time point of the observed stage; 2) if a larva advanced one or more stages during an observation time interval (i.e. 24 or 48 hours), that larva was assumed to have progressed through all intervening stages at a constant rate (calculated as the time interval divided by the stage differential); and 3) if a larva did not progress at all during a 48 hour observation interval, the total length of that stage for that larva could not be estimated, so we scored the length of that stage for such larvae as 72 hours. We consider this maximum length of 72 hours to be quite conservative, based on our observations on cohorts of larvae that indicate that none of our skeletogenic stages are of that duration at 14-16°C (data not shown).
For example, the length of stages for a hypothetical larva that began the experiment (0 hrs) at Stage 6, progressed to Stage 7 at 24 hours, and then Stage 9 at 48 hours would be computed as follows. Based on assumption 1: the larva was at the midpoint of Stage 6 at 0 hrs, of Stage 7 at 24 hours and Stage 9 at 48 hours. Sometime during the second 24 hours, the larva entered and exited Stage 8. Based on assumption 2: the lengths of Stage 7, 8 and 9 are assumed to be equal, and thus 12 hours each (time interval = 24 hours, stage differential = 2 stages; therefore 24/2 = 12 hr). This means that Stage 7 is assumed to have begun 18 hours into the experiment, Stage 8 at 30 hours, and Stage 9 at 42 hours). And, finally, Stage 6 – which was at its midpoint at 0 hrs (see assumption 1) – is 36 hours for this larva (it is assumed that Stage 6 for this larva began 18 hours before the onset of the experiment and finished when Stage 7 is assumed to have begun: namely, 18 hours after the start of the experiment).
We calculated means, standard errors of the mean and the 95% confidence intervals for the lengths of each stage at each temperature. Note that we could not estimate the length of Stage 10 in our study, since we calculated stage lengths based on progression into the subsequent stage, and there is no Stage 11 in our scheme.
In order to confirm that our live mounting technique of larvae did not affect the developmental progression of larvae, we setup a separate experiment with a total of 24 larvae in February 2014 at UOG. 12 larvae were randomly assigned to a mounting treatment. Specifically, we mounted these 12 larvae as described above, staged them and placed them back into individual wells as described above. The other 12 larvae were placed directly into wells without mounting. After 48 hours, all 24 larvae were staged, and we compared the distribution of stages in the two sets of larvae (manipulated vs. non-manipulated).
Purple urchin larvae have two types of spines: 6-sided “adult” spines and 4-sided “juvenile” spines (see ) for review, and for a listing of alternative names in the literature for these spine types). The latter are so named as they are juvenile-specific, and are not found in urchin adults. Due to their greater numbers, ease of scoring and regular positions within the growing rudiment, we decided to focus exclusively on the adult spines for this particular analysis. Using the same dataset described above (Guelph experiment), we counted the number of cross hatches in the first five adult spines that we could observe in each larva that had reached Stage 8. We used these data to calculate mean and maximum number of cross hatches for each larva.
We calculated growth rates of the adult spines within the first and second 24 hour period of observations to get a rough estimate of spine growth in S. purpuratus larvae. For this analysis we counted the regular cross bars that occur along the length of the growing adult spines (see Figure 5A) within the rudiment during the final week or so of development before settlement; since these bars occur at regular intervals along the length of the growing spine (data not shown), the number of cross bars (=cross hatches) is a convenient way to estimate spine length. For each larva in the Guelph experiment, that was at or beyond Stage 8 (see Table 2), we counted the number of complete cross hatches (i.e. with no gap) in the first five adult spines that we identified while examining larvae at 0, 24 and 48 hours. We then calculated the mean (Avg) and maximum (Max) number of cross hatches for the five counted spines in each of these larvae, and used these values to compare mean rates of spine growth (for both Avg and Max) during the first and second 24 hour periods of observation.
Note that adult spines are hexagonal around the long axis of the spine, and thus have cross bars along all six edges of the hexagon (i.e., along all six faces of the spine), and that the number of cross hatches along each of these six faces is not always consistent within a spine (e.g., see Figure 5A). We counted the maximum number of cross hatches visible on any of the six faces. For example, in a spine where the six faces had 2, 2, 3, 2, 2 and 3 cross hatches, it would receive a score of 3 cross hatches. In another example, if only one complete cross hatch appears on as few as one face, that spine would receive a score of 1 cross hatch. We analyzed these data using a one tailed t-test for 24 and 48 h in order to assess whether the rate of addition of cross hatches was different during the first and second 24 hours of our experiment.
Comparison to other staging schemes
In Tables 1 & 2, we present comparisons between our staging scheme and the staging schemes for S. purpuratus and 3 other echinoids, as presented by Smith et al.  and Chino et al.  respectively. For these comparisons, we examined the images, drawings and descriptions presented by these authors, and compared them to the new soft tissue and skeletogenic stages that we present here.
In Guelph, we imaged larvae on a Nikon Ti microscope equipped with a motorized stage. Images of larvae were taken at each time point and, if necessary, multiple views were captured of juvenile skeletons. In Seattle, we imaged only a select subset of larvae to illustrate certain stages, using a Nikon Coolpix 990 camera mounted on a Leitz Wetzlar Ortholux microscope. We captured additional images to illustrate some stages in Figure 1 using a Zeiss Axio Imager Z1 microscope at HMS. All three of these systems were equipped with DIC optics and cross-polarized light. All image are oriented with posterior to the left.
Research presented here does not require any approval and therefore complies with all necessary regulations.